Neuroscience of Consciousness, 2016, 1–9
doi: 10.1093/nc/niw013
Rapid Communication
Subliminal messages exert long-term effects on
decision-making
Simon Ruch*, Marc Alain Züst, and Katharina Henke
Department of Psychology, University of Bern, Switzerland and Center for Cognition, Learning and Memory,
University of Bern, Switzerland
*Correspondence address. Department of Psychology, University of Bern, Fabrikstrasse 8, 3012 Bern, Switzerland. Tel: +41 (0)31 631 40 11; E-mail:
[email protected]
Abstract
Subliminal manipulation is often considered harmless because its effects typically decay within a second. So far, subliminal
long-term effects on behavior were only observed in studies which repeatedly presented highly familiar information such
as single words. These studies suggest that subliminal messages are only slowly stored and might not be stored at all if they
provide novel, unfamiliar information. We speculated that subliminal messages might affect delayed decision-making especially if messages contain several pieces of novel information that must be relationally bound in long-term memory.
Relational binding engages the hippocampal memory system, which can rapidly encode and durably store novel relations.
Here, we hypothesized that subliminally presented stimulus pairs would be relationally processed influencing the direction
of delayed conscious decisions. In experiment 1, subliminal face–occupation pairs affected conscious decisions about the
income of these individuals almost half an hour later. In experiment 2, subliminal presentation of vocabulary of a foreign
language enabled participants to later decide whether these foreign words are presented with correct or incorrect translations. Subliminal influence did not significantly decay if probed after 25 versus 15 min. This is unprecedented evidence of
the longevity and impact of subliminal messages on conscious, rational decision-making.
Key words: subliminal; decision-making; unconscious processing; long-term memory; hippocampus; information integration
Introduction
Subliminal messages exert diverse influences on our thoughts
and our behavior (van Gaal et al., 2012; Hassin, 2013). Subliminal
stimuli can facilitate conscious processing of related information
(Van den Bussche et al., 2009), change our current mood (Monahan
et al., 2000), boost our motivation (Aarts et al., 2008), and can even
alter our political attitudes and voting intentions (Hassin et al.,
2007; Weinberger and Westen, 2008). With such a broad impact,
subliminally planted information might have the potential to alter our decisions in everyday situations such as voting.
In order to influence decision-making in real-life situations,
subliminal messages must be stored for long-term after only a
few exposures, e.g. after a single confrontation with a subliminal TV advert. Furthermore, messages must be stored even if
they contain complex relational information that requires semantic integration, such as “politician X will lower the taxes.”
For subliminal manipulation to be effective, humans thus have
to be able to semantically integrate and rapidly store unconscious pieces of novel information into long-lasting associative
memories that can be retrieved if relevant to the context of a
later decision.
The processes which allow novel information to shape
subsequent decisions are generally thought to depend on consciousness – be it the integration of novel information into abstract mental representations (e.g. Tononi, 2004), rapid
encoding of these representations into long-term memory (e.g.
Shanks, 2010), or the use of these representations to make informed decisions (e.g. Newell and Shanks, 2014). However,
Received: 9 December 2015; Revised: 1 July 2016. Accepted: 22 July 2016
C The Author 2016. Published by Oxford University Press.
V
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/
licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited.
For commercial re-use, please contact [email protected]
1
2
| Ruch et al.
growing evidence indicates that the human unconscious can
perform various high-level cognitive functions (van Gaal et al.,
2012; Hassin, 2013) that might allow decision processes to benefit from subliminal messages.
Several studies reported that subliminally planted information can be semantically integrated outside conscious awareness
(for a detailed review, see Mudrik et al., 2014). Indeed, humans can
detect incompatible object-background configurations in subliminal scenes (Mudrik and Koch, 2013), solve subliminally presented
arithmetic problems (Garcıa-Orza et al., 2009; Van Opstal et al.,
2011; Sklar et al., 2012; Karpinski et al., 2016), and draw inferences
from subliminal picture sequences and word pairs (Kawakami
and Yoshida, 2015 and Reber and Henke, 2012, respectively).
Subliminal stimulation was further found to nonconsciously
shape decision-making – at least if masked stimuli consisted of
single familiar items that required little integration. For example,
priming studies reported that subliminal primes not only facilitate
correct responses to related targets in a classification task, but
also bias responses in “free choice” tasks in which participants
can freely decide between response alternatives (Schlaghecken
and Eimer, 2004; Klapp and Haas, 2005; Kiesel et al., 2006;
Parkinson and Haggard, 2014; Ocampo, 2015). Similarly, studies on
subliminal persuasion suggested that repeated subliminal exposure to brand names (e.g. “Lipton Ice Tea”) or goal-relevant words
(e.g. “thirst” in thirsty participants) can bias participants’ product
choices (Karremans et al., 2006; Bermeitinger et al., 2009;
Verwijmeren et al., 2011, 2013) or reinforce a certain behavior (e.g.
to drink, see Strahan et al., 2002). Thus, there is ample evidence
that subliminal messages can be integrated unconsciously and
can influence decisions and choices.
Whether subliminally presented information is stored in
long-term memory to guide delayed decisions is vastly unknown.
So far, studies on information integration and decision-making
only assessed immediate influences of subliminal stimulation.
Priming studies which assessed the longevity of subliminal influences usually reported that behavioral effects of masked primes
decay within 1 s (e.g. Forster et al., 1990; Ferrand, 1996; Greenwald
et al., 1996). This suggests that subliminal information is not
stored and thus cannot affect delayed decisions. Yet, some studies reported that longer lasting subliminal influences on behavior
are possible under certain conditions. In many of these studies,
the same subliminal messages were presented multiple times
(Lowery et al., 2007; Capa et al., 2011; Levy et al., 2014; Farooqui
and Manly, 2015). Furthermore, participants were often informed
about the presence of subliminal stimuli and were provided conscious feedback or rewards after each subliminal message (Aarts
et al., 2008; Capa et al., 2011; Farooqui and Manly, 2015) or were
asked to consciously detect or classify each subliminal event
(Gaillard et al., 2007; Chen et al., 2009; Chong et al., 2014). These
studies thus suggest that subliminal messages are only slowly
stored and are only retained if subjects have the explicit intention to process the hidden events. Importantly, most studies
used familiar information such as single words as subliminal
stimuli to prime subsequent conscious processing of this information (Gaillard et al., 2007; Chen et al., 2009; Chong et al., 2014)
or to prime a specific goal (e.g. to perform well on a test), intention, or stereotype (Lowery et al., 2007; Capa et al., 2011; Levy
et al., 2014). This suggests that subliminal long-term effects are
achieved only if familiar information is presented but not if novel
relational information has to be learned.
We asked if humans can rapidly integrate and store novel
relational information (e.g. “person X is a manager”, see Fig. 1a)
from subliminal messages for later use in a decision-making
situation (e.g. “guess the income of X”). We speculate that
subliminal messages should be stored especially if they consist
of multiple items that require relational processing. Relational
binding calls upon the hippocampal memory system, which
can rapidly store novel relations for long term (Henke et al.,
1997; Holdstock et al., 2002; Harand et al., 2012). Traditional
views hold that hippocampus is only involved in the encoding
and retrieval of consciously perceived information (Moscovitch,
1995; Squire and Zola, 1996; Tulving, 2002), and that associative
learning outside conscious awareness is unlikely (Shanks,
2010). However, growing evidence suggests that hippocampus
operates independently of consciousness and that nonconscious relational learning is humanly feasible (for reviews, see
e.g. Reder et al., 2009; Henke, 2010; Dew and Cabeza, 2011;
Hannula and Greene, 2012; Olsen et al., 2012). Indeed, hippocampus was found to mediate implicit learning (Chun and
Phelps, 1999; Greene et al., 2006; Negash et al., 2015) and
retrieval (Greene et al., 2007; Hannula and Ranganath, 2009;
Addante, 2015; Reber et al., 2016) of relational information between visible stimuli. Most importantly, hippocampus was also
found to be involved in the encoding and retrieval of subliminally presented stimulus pairs (Henke et al., 2003; Reber et al.,
2012; Duss et al., 2014; Züst et al., 2015).
We ran two experiments to test whether subliminal stimulus pairs would affect delayed decision-making (Figure 1a).
In experiment 1, we tested whether subliminal presentations of
face–occupation pairs would guide later conscious decisions
about the income of the same faces. We assessed the longevity
of subliminal influence by measuring its decay across delays of
15–25 min. In experiment 2, we assessed if new vocabulary of a
foreign language is acquired subliminally affecting later lexicalsemantic decisions on the same foreign words. Both experiments were composed of an encoding phase, an encoding-test
interval that was filled with rest, and a test phase (Fig. 1a). The
decision task given in the test phase constituted an indirect (implicit) memory test. Importantly, participants were kept naı̈ve
regarding subliminal stimulation until the end of the experiment. This allowed us to assess subliminal influences in the absence of any explicit intention to process the hidden messages.
Method
General procedure
Unless indicated otherwise, experimental procedures were
identical for both experiments. We used a well-established paradigm to render stimuli invisible (Duss et al., 2011). Each stimulus was presented 12 times for 17 ms during one unconscious
encoding episode of 6 s duration (see Supplementary Fig. 1).
Stimuli were preceded and followed by visual noise masks
(sandwich masking). In experiment 1, 40 face–word pairs were
encoded in 40 unconscious encoding episodes. In experiment 2,
each unconscious encoding episode contained a pseudoword–
word pair that was presented twice in nonadjacent repetitions,
yielding a total of 48 encoding episodes for 24 pseudoword–
word pairs. Stimuli and masks were embedded in an attention
task which allowed us to direct participants’ focal attention to
the screen without disclosing the presence of subliminal
messages (see Supplementary Fig. 1). The task required participants to respond to target screens that appeared at random
times once in every encoding episode (see Supplementary Fig.
1). Mean hit rate to targets was >85% in both experiments.
Following the attention task, participants rested for 15, 20, or
25 min (depending on experiment and condition) before performing the decision tasks. The attention task and the decision
Subliminal messages—effects on decision-making |
(a)
3
(b)
Figure 1. (a) Design of experiments 1 and 2. Subliminal presentation of face–occupation pairs (Experiment 1) or pairs consisting of pseudowords
and their translations (Experiment 2) altered participants’ decisions when later asked to rate the income of faces or to identify correct (matching) translations of pseudowords. (b) Results. Two-tailed t-tests indicated that decision accuracy was significantly above the 50% chance level
(mean accuracy with 95% confidence interval). *P < 0.05, **P < 0.01.
task were briefly practiced at the beginning of the experiment
using a few stimuli that were not used later on.
Following the main experiments, a structured interview was
performed to assess subjective awareness for subliminal information. Using funneled questions, we first asked participants
whether they had seen any unexpected stimuli during the attention task or had suspected the presence of any hidden information, and then asked more specifically whether they had
seen hidden faces or words. We then informed participants of
the subliminal stimulation and administered forced-choice
tests to assess awareness of subliminal stimuli objectively.
Masking paradigm and attention task in these awareness tests
were the same as in the main experiments. However, awareness was tested trial-by-trial: following each encoding episode
of a new stimulus pair, the attention task was interrupted and
participants were asked to provide information about the subliminal stimuli. We expected participants to perform at chance
level in these tests if masking rendered stimuli truly subliminal.
Experiments were approved by the local ethics committee.
Written semi-informed consent was obtained from all participants before experimentation. The information about subliminal stimulation was provided following the experiment.
Participants
We recruited participants with normal or corrected-to-normal
visual acuity. In experiment 1, two participants were excluded
post hoc due to errors in data logging. Of the remaining 46 participants (19 men and 27 women, mean age ¼ 36.0), 23 were tested
with a 15 min delay between subliminal encoding and decisionmaking, and 23 with a 25 min delay. In experiment 2, two participants were excluded post hoc because they had failed to report
their decisions within the response time window. The remaining
sample consisted of 23 women and 11 men (mean age ¼ 22.8).
Experiment 1
During the encoding phase, we presented 10 faces combined
with written high-wage occupations and 10 faces with lowwage occupations (Fig. 1a). Twenty additional faces were paired
with consonant strings and presented subliminally in a control
condition (not reported). Each stimulus pair was presented in
just one subliminal encoding episode, which comprised 12 stimulus repetitions within 6 s. The encoding-test interval spanned
15 or 25 min. In the test phase, the former subliminal faces were
re-presented for 5 s for conscious inspection with the instruction to decide whether an individual would earn a high or low
income.
Objective awareness tests
We administered two forced-choice tests to assess awareness
of subliminal faces and occupations objectively. In each test, we
presented 40 novel face–occupation pairs subliminally using the
same attention task as in the main experiment. Immediately
following the encoding of a novel pair in a subliminal encoding
episode, the attention task was interrupted and participants
were interrogated regarding the just presented face or occupation. In the face-awareness test, the just subliminally presented
face and a novel face were presented side by side for conscious
inspection. Participants were asked to select the face that they
thought was just flashed subliminally. In the occupationawareness test, the text “Income?” was presented which signaled participants to name the income (“high” vs. “low”) that they
thought would the just subliminally flashed occupation yield.
After giving their response, the attention task was continued
and a new face–occupation pair was presented subliminally.
Participants took either 40 face-awareness trials followed by 40
occupation awareness trials or vice versa.
Material
We used 160 grayscale images of male faces that were given average income ratings in a pilot study. Images were equalized regarding luminance and contrast and were assigned to 16 lists of
10 faces each. Lists were comparable with respect to income,
age, facial hair, and emotional facial expression. Lists were
counterbalanced over experimental conditions such that each
face was presented an equal number of times with a high- and
a low-wage occupation and with a consonant string. All faces
were also rotated into the two awareness tests. Stimuli were
presented an equal number of times in the main experiment,
and the awareness tests of faces and occupations. Hence, results in the experiment and the awareness test derived from
the same stimulus material (presented to different
participants).
We used 10 typical high- and 10 low-wage occupation words
that were similar regarding mean logarithmic word frequency
4
| Ruch et al.
(Leipzig Corpora Collection, http://corpora.uni-leipzig.de/) and
character count.
Faces were displayed at the center of the screen in front of a
dark gray background. Words were presented below the faces in
light gray in a sans-serif font. Stimuli were delivered at a visual
angle of 15 using a Digital Light Processing (DLP) projector with
a 60 Hz refresh rate.
Experiment 2
Participants encoded 24 combinations of written pseudowords
(fictitious foreign language) and German words (fictitious translations). We presented these pseudoword–word combinations
subliminally for unconscious semantic relational encoding using the same paradigm as in experiment 1. There were two
nonadjacent subliminal encoding episodes (each comprising 12
stimulus repetitions within 6 s) per word pair within a randomized sequence of subliminal encoding episodes. The encodingtest interval spanned 20 min. At test, we re-presented the same
foreign words for 5 s for conscious inspection. Each foreign
word was shown besides a German word that was a synonym
to the subliminal German translation of either this or another
foreign word. Synonyms were used to test for semantic rather
than perceptual relational retrieval. Half of foreign words presented at test were recombined to break the encoded semantic
relation (incorrect translations) and half were combined to keep
the semantic relation from encoding to test (correct translation).
Participants were instructed to decide whether or not a presented foreign word and the German translation word fit
together (match/mismatch decision).
Objective awareness test
We administered one forced-choice test to assess awareness
of subliminal words objectively. Participants encoded 24 (novel)
pseudoword–German word pairs. Following each subliminal
encoding episode, a probe word was displayed for conscious inspection for participants to decide whether or not the probe was
a synonym to the just presented subliminal German word. In
half the trials, the probe was a valid synonym to the subliminally presented German word, in the other half the probe was
an unrelated foil. Hence, we assessed the semantic processing
of subliminal words, as in the main experiment.
Material
We created 48 word triplets consisting of a pseudoword, its
German translation used for subliminal encoding, and a synonym to the German translation used for the decision task
(see Fig. 1). German synonym pairs were gathered from Open
Thesaurus (http://www.openthesaurus.de/). Two-syllabic pronounceable pseudowords were created using German and Dutch
syllables provided by the Celex database (http://celex.mpi.nl/).
The 48 encoding stimuli were assigned to four lists of 12 items
each with equal distributions of word lengths, pronounceability, concreteness, animatedness, and logarithmic frequency of
appearance (drawn from Leipzig Corpora Collection, http://cor
pora.uni-leipzig.de/). Assignment of German translations to
pseudowords was randomized anew for each list and each participant to reduce potential bias resulting from particular combinations. Two lists were used for the main experiment and the
other two lists for the awareness test; this assignment was counterbalanced over participants. Twelve further German words
were used as foils in the awareness test. Pseudowords were displayed in the left and German words in the right visual field
of participants. Words were presented for subliminal encoding
using the same masking paradigm and psychophysical conditions as applied in experiment 1.
Results
Experiment 1
In experiment 1, we tested whether subliminal presentations of
face–occupation pairs would influence later classification of the
same faces. We hypothesized that participants would rate individuals who had earlier been flashed with a high-wage or
low-wage occupation as high-income or low-income earners,
respectively. A two-tailed t-test on the mean decision accuracy
confirmed this hypothesis. Mean decision accuracy was 53.97%
(95% CI [51.25, 56.96]), which exceeded the chance level of 50%
(t(45) ¼ 2.94, P ¼ 0.005; r ¼ 0.40) (Fig. 1b). Mean decision accuracy
was not significantly smaller after 25 compared to 15 min
(53.04% vs. 54.90%; two-tailed t(44) ¼ 0.68, P ¼ 0.50; r ¼ 0.10),
which indicates that the effect of subliminal encoding persisted
through both retention intervals at equal strength. Hence, new
semantic associations were stored for long term affecting the
direction of decision-making almost half an hour later.
An interview that was administered after the main experiment to assess subjective awareness of the masked stimuli suggested that none of the participants had seen the subliminal
faces or words during encoding. Two objective awareness tests
further indicated that masked stimuli could not be consciously
perceived. Participants performed at chance level of 50% if
asked which one of two faces had just been presented subliminally (mean recognition accuracy ¼ 48.81%, 95% CI [46.36, 51.27],
missing data of one participant due to data loss) or if a high-or
low-wage occupation had been presented (mean recognition
accuracy ¼ 49.06%, 95% CI [46.90, 51.23]). The t-tests against
chance level were not significant (awareness for faces:
t(43) ¼ 0.98, P ¼ 0.34, r ¼ 0.15; awareness for occupations:
t(44) ¼ 0.90, P ¼ 0.37, r ¼ 0.13), suggesting that participants were
consciously unaware of the subliminal stimuli. However, the
nonsignificant tests do not reveal whether participants were
truly unaware (i.e. whether H0 can be accepted), or whether our
tests were not sensitive enough to detect awareness. We, therefore, calculated Bayes factors for the two t-tests following the
recommendations provided in Dienes (2008, 2014). Bayes factors
(BF) indicate the relative strength of two hypotheses H1 and H0
with a BF(H1/H0) ¼ 1 suggesting that the data are inconclusive
and favor neither hypothesis, and BFs > 3 or < 1/3 suggesting
substantial evidence for H1 or H0, respectively. Assuming that
awareness of subliminal stimuli should yield recognition accuracy that is comparable to performance in the decision task
(3.97% above chance level), we chose a half-normal prior distribution with a mode of 0 and a standard deviation of 3.97% to
calculate BF (Dienes, 2014). The resulting BF for the face-awareness test (mean accuracy ¼ 1.19% below chance level,
SE ¼ 1.21%) and the occupation-awareness test (mean accuracy ¼ 0.94% below chance, SE ¼ 1.04%) were 0.16 and 0.14, respectively. Both factors were below 1/3, indicating substantial
evidence for the null hypothesis, i.e. that participants were not
aware of the masked stimuli. We further performed a multiple
regression analysis to assess whether potential awareness of
faces or occupations predicted decision accuracy in the main
experiment, and to estimate whether accuracy remained above
chance level if the theoretical awareness in both tests is 0 (i.e.,
whether the intercept is significant). This method was brought
forward by Greenwald et al. (1995). Although subject to criticism
(e.g. Miller, 2000), this method is used widely to assess whether
Subliminal messages—effects on decision-making |
subliminal influences on behavior are independent of stimulus
awareness. Neither potential awareness of faces (b ¼ 0.16,
t(41) ¼ 0.88, P ¼ 0.383) nor potential awareness of occupations
(b ¼ 0.24, t(41) ¼ 1.2, P ¼ 0.23) predicted decision accuracy in
the main experiment. Decision accuracy in the main experiment remained above chance level when we controlled for selection accuracy in the two awareness tests (intercept of the
regression with two predictors; t(41) ¼ 2.93, P < 0.01, r ¼ 0.42). In a
final step, we excluded those participants from the analysis of
the data from the main experiment, who tended to perform
above or below chance level in either of the two awareness tests
(12 participants exhibited an awareness score with a binomial
probability of P < 20%). Classification accuracy in the main experiment remained above chance level (54.12%, 95% CI [50.72,
57.51], t(33) ¼ 2.468, P ¼ 0.019, r ¼ 0.39) with these 12 participants
removed. We, therefore, conclude that there was no awareness
of encoding stimuli in the main experiment and that long-term
effects derived from unconscious processing alone.
Experiment 2
In experiment 2, we tested for long-term effects using a subliminal vocabulary acquisition task. We hypothesized that new vocabulary would be encoded subliminally influencing delayed
lexical–semantic decisions on the same foreign words when
they were presented visibly. At test, participants were instructed to decide whether or not a given foreign word and a
German translation word fit together (match/mismatch decision). On average, 52.7% (95% CI [50.1, 55.2]) (Fig. 1b) of synonyms were classified accurately, which exceeded chance level
(two-tailed t-test: t(33) ¼ 2.128; P ¼ 0.041; r ¼ 0.35). Hence, the
meaning of subliminal German words was decoded, linked to
foreign words, and stored to influence lexical–semantic decisions 20 min following subliminal encoding.
An interview administered following the main experiment
assessed the subjective awareness of the masked stimuli.
Participants’ responses indicated that none had consciously
perceived subliminal words or letters. An objective awareness
was used to corroborate the subjective reports. When asked
whether a consciously displayed German word represents a
synonym to a just subliminally flashed German word, participants decided correctly in 51.3% of cases (95% CI [48.9, 53.8]).
This performance did not significantly exceed chance level of
50% (t(33) ¼ 1.121, two-tailed P ¼ 0.27; r ¼ 0.19). To validate
whether performance was truly at chance level, we again calculated the Bayes factor for the t-test as suggested by Dienes
(2008, 2014). Assuming that awareness for subliminal words
should yield a recognition accuracy comparable to performance
in the decision task (2.7% above chance level), we chose a
half-normal prior distribution with a mode of 0 and a standard
deviation of 2.7% to estimate the BF (Dienes, 2014). The resulting
factor for the awareness test (mean accuracy ¼ 1.3% above
chance level, SE ¼ 1.16%) was 1.14, which suggests that the test
was not sensitive enough to reject stimulus awareness.
Whether participants were truly unaware of the subliminal
words thus remains elusive. However, further analyses suggested that subliminal long-term influences on decision-making did not benefit from stimulus awareness. First of all, a
regression analysis (Greenwald et al., 1995) indicated that accuracy in the awareness test did not predict decision accuracy in
the main experiment (b ¼ 0.038, t(32) ¼ 0.216, P ¼ 0.83), which
was above chance level when selection accuracy on the awareness test was 0 (intercept of regression; t(31) ¼ 2.052, P ¼ 0.045,
r ¼ 0.34). Furthermore, even if those two participants, who
5
tended to perform above or below chance level in the awareness
test (binomial probability of P < 20%), were excluded from the
analysis of the data from the main experiment, classification
accuracy remained above chance level (52.6%, 95% CI [50.1,
55.1], t(31) ¼ 2.142, P ¼ 0.040, r ¼ 0.36). Although the Bayes analysis yielded an inconclusive result regarding stimulus awareness
in the awareness test, the regression analysis suggests that
long-term effects on decision-making derived from unconscious
processes.
Discussion
To summarize, subliminal verbal and nonverbal item pairs
influenced participants’ conscious, deliberate decisions almost
half an hour following subliminal stimulation. In experiment 1,
subliminal face–occupation pairs influenced conscious decisions on the income of the same individuals after delays of 15
and 25 min. In experiment 2, subliminal presentation of foreign language vocabulary influenced participants’ decisions on
correct/incorrect word translations after a delay of 20 min.
These findings provide unprecedented evidence of a considerable longevity of subliminal effects on intentional behavior
such as decision-making.
So far, most investigators who addressed the longevity of
subliminal priming reported that subliminal effects on behavior
would fade within 1 s (e.g. Greenwald et al., 1996), which suggests that subliminal messages leave no long-term memory
traces and therefore cannot influence delayed decisions. The
few studies which reported longer lasting effects used familiar
stimuli that were repeatedly presented to participants who
were informed about the presence or purpose of subliminal
events (e.g. Chong et al., 2014; Levy et al., 2014). These studies
suggest that subliminal information may be stored for long term
if it is familiar, if many subliminal exposures are provided for a
slow, incremental encoding process, and if subjects explicitly
intent to process the subliminal information. Whether novel
subliminal information can be integrated and stored following a
single exposure to naı̈ve subjects remains unclear. Even less
clear is whether a subliminal one-time processing has sufficient
strength to influence delayed decision-making. We found that
participants, who were unaware of the presence and purpose of
subliminal information, successfully processed subliminal face–
word and nonword–word pairs and formed lasting unconscious
semantic relational memories based on only one or two subliminal encoding episodes (each comprising 12 adjacent subliminal
stimulus repetitions).
We speculate that the reported rapid encoding and longterm retention of subliminal information owes to the type of
stimuli and the memory system these stimuli called upon. As
mentioned above, conclusions regarding the longevity of subliminal priming rested on the use of familiar single-item stimuli
such as words (e.g. Forster et al., 1990; Ferrand, 1996; Greenwald
et al., 1996). It is known that the processing of single items engages neocortex (Henke et al., 1997; Duss et al., 2014), which
forms long-term memory traces rather slowly over many learning trials (McClelland et al., 1995). This might explain why subliminal influences were found to be short-lived or to build up
slowly. Here, we used multi-item displays that contained both
novel (unfamiliar faces or foreign words) and familiar (occupation words or German translation words) pieces of information.
Relational encoding of multiple stimuli engages hippocampus
(Henke et al., 1997; Duss et al., 2014), which learns rapidly and
stores relations for long term due to its exceptional plasticity
(McClelland et al., 1995). Subliminal relational encoding and
6
| Ruch et al.
retrieval have been associated reliably with hippocampal activity changes (Henke et al., 2003; Reber et al., 2012; Züst et al., 2015).
Importantly, hippocampal damage abolished both supraliminal
(conscious) and subliminal (unconscious) relational binding but
left subliminal single-item priming intact (Duss et al., 2014). As
encoding-test intervals spanned a few minutes or less in these
previous neuroimaging experiments, they were uninformative
regarding longer term effects of subliminal stimuli on behavior.
Using similar stimuli and the same masking technique, we now
demonstrate subliminal effects on decisions delayed by
15–25 min. Because subliminal influences did not noticeably decay from the delay of 15–25 min in experiment 1, we speculate
that subliminally planted information might affect decisionmaking even at longer intervals.
The long-lasting influence of subliminal stimulus pairs evinces nonconscious relational learning, but the experimental paradigms employed in this study do not pin down the exact
nature of the unconsciously formed memories. Although we assume that subliminal stimulus pairs yield semantically precise
unconscious relational memories (“person X is a manager”,
“gumpel means dog”) that are later reactivated to support decision-making, our experimental setup cannot rule out the possibility that participants had formed associations between faces/
foreign words and broad semantic or affective categories (e.g.
“person X as a manager must be wealthy”, “I despise gumpel because I fear dogs”). Such fuzzy semantic/affective associations
could suffice to guide subsequent decisions (“X has a high income because he is wealthy”, “hound is a valid translation of
gumpel because I despise both”). But we would like to point out
that the results of earlier investigations on subliminal encoding
and long-term memory formation demonstrated unequivocally
that subliminal words are understood with high semantic
precision. In Duss et al. (2011), subliminal presentations of
face–occupation pairs influenced participants’ subsequent conscious classifications of the same faces when various semantic
dimensions were offered, namely regularity of income, length
of education, and creativity of work. Faces encoded with an artistic instead of an academic occupation (e.g. “actor” vs.
“lawyer”) were later classified as generating an irregular income, coming from a shorter school education, and performing
creative work. These distinct influences of subliminal information on diverse semantic classifications suggest a precise rather
than diffuse lexical–semantic word analysis. Further evidence
for precise subliminal encoding is provided by our studies on
unconscious relational inference (Reber and Henke, 2012; Reber
et al., 2012; Henke et al., 2013). In these studies, overlapping
subliminal word pairs such as “winter-red” and “red-computer”
(A–B, B–C) were presented apart in time and were nevertheless
integrated semantically to influence delayed judgments
regarding the semantic relatedness of A and C, like “wintercomputer”. The successful relational integration in this
subliminal paradigm cannot be explained by affective or fuzzy
semantic word priming.
Doubts have been raised recently as to whether experimenters had adequately assessed stimulus awareness and correctly estimated influences of subliminal stimuli on behavior
(Newell and Shanks, 2014; Hesselmann and Moors, 2015).
To avert such suspicion, we assessed stimulus awareness following the main experiments using objective awareness tests
that had the same statistical power as our decision tasks. The
objective awareness tests confirmed that the subliminal stimulation paradigm rendered stimuli largely (experiment 2) or completely (experiment 1) imperceptible to the conscious mind, and
that the observed long-term effects of subliminal stimulation
were independent of stimulus awareness. Because we did not
assess awareness of each subliminal stimulus immediately following the respective encoding episode in the main experiments, we cannot rule out the possibility that participants were
briefly aware of some of the subliminal images. However, none
of the participants reported to have noticed the presence of hidden or masked information during the main experiments or
during the objective awareness tests. Hence, all available data
suggest that the reported long-term influences of subliminal
stimuli were independent of consciousness.
The finding that subliminally processed information is rapidly integrated and stored to guide delayed decisions challenges
prevailing views of the cognitive function of consciousness.
Consciousness is usually considered a precondition for successful information integration (Tononi, 2004; Mudrik et al., 2014),
relational learning (Shanks, 2010), and decision-making
(Bettman et al., 1998; Simonson, 2005; Newell and Shanks, 2014).
However, evidence is accumulating that these notions of consciousness need revision (Dijksterhuis and Nordgren, 2006;
Nordgren, 2006; Reder et al., 2009; Henke, 2010; Dew and Cabeza,
2011; Olsen et al., 2012; Hannula and Greene, 2012; van Gaal
et al., 2012; Hassin, 2013). Indeed, unconscious integration of different semantic concepts and of temporally or spatially distributed percepts was reported not only for stimuli that were
rendered subliminal using visual masking but also for stimuli
made invisible using continuous flash suppression (Mudrik
et al., 2011; Sklar et al., 2012; Vlassova et al., 2014; Bergström and
Eriksson, 2015; Karpinski et al., 2016) or visual crowding (Atas
et al., 2013). Even during the unconsciousness of deep sleep,
words, sounds, and odors were found to be integrated (Ruby
et al., 2008; Daltrozzo, et al., 2012) and stored in long-term memory (Arzi et al., 2012, 2014; Ruch et al., 2014) to modulate behavior
following waking. More evidence for the feasibility of unconscious relational integration is provided by social psychology:
decisions, which require the consideration, weighing and integration of large amounts of (supraliminally provided) information, were better following unconscious deliberation than
conscious reasoning (e.g. Dijksterhuis et al., 2006; Mealor and
Dienes, 2012; Abadie et al., 2013). This ‘deliberation without attention’ effect suggests that our conscious decisions are vitally
influenced by nonconscious processes (but see e.g. Newell and
Shanks, 2014; Nieuwenstein et al., 2015; and Vadillo et al., 2015
for critical reviews).
In sum, our findings add to a growing body of evidence suggesting that subliminal messages can be used to change our
thoughts, attitudes, emotions, and actions (van Gaal et al., 2012;
Hassin, 2013). Subliminal stimulation was shown to influence
current moods (Monahan et al., 2000), political attitudes (Hassin
et al., 2007; Weinberger and Westen, 2008), intentions (Hassin
et al., 2007), choices and decisions (Bermeitinger et al., 2009), and
cognitive strategies (Lau and Passingham, 2007; Reuss et al.,
2011). However, little is known about the longevity of these subliminal influences because these studies were focused mainly
on immediate effects of subliminal stimulation. Here, we demonstrate that a few exposures to novel subliminal information
is sufficient to influence delayed decision-making. The surprising impact of subliminal messages on rational, intentional, conscious behavior lends subliminal protocols to practical
applications, of which advertising is just one example.
Supplementary data
Supplementary data is available at Neuroscience of
Consciousness Journal online.
Subliminal messages—effects on decision-making |
Acknowledgments
This work was supported by Swiss National Science
Foundation
Grants
K-13K1-119953
to
K.H.
and
P0BEP1_148941-1 to M.A.Z.
Behavioral data are available on request.
References
Aarts H, Custers R, Marien H. Preparing and motivating behavior
outside of awareness. Science 2008;319:1639. http://doi.org/10.
1126/science.1150432.
Abadie M, Waroquier L, Terrier P. Gist memory in the
unconscious-thought effect. Psychol Sci 2013;24:1253–9. http://
doi.org/10.1177/0956797612470958.
Addante RJ. A critical role of the human hippocampus in an electrophysiological measure of implicit memory. NeuroImage
2015;109:515–28. http://doi.org/10.1016/j.neuroimage.2014.12.069.
Arzi A, Holtzman Y, Samnon P, et al. Olfactory aversive conditioning during sleep reduces cigarette-smoking behavior. J
Neurosci 2014;34:15382–93. http://doi.org/10.1523/JNEUROSCI.
2291-14.2014.
Arzi A, Shedlesky L, Ben-Shaul M, et al. Humans can learn new
information during sleep. Nat Neurosci 2012;15:1460–5. http://
doi.org/10.1038/nn.3193
Atas A, Faivre N, Timmermans B, et al. Nonconscious learning
from crowded sequences. Psychol Sci 2013. http://doi.org/10.
1177/0956797613499591.
Bergström F, Eriksson J. The conjunction of non-consciously perceived object identity and spatial position can be retained during a visual short-term memory task. Conscious Res 2015;1470.
http://doi.org/10.3389/fpsyg.2015.01470
Bermeitinger C, Goelz R, Johr N, et al. The hidden persuaders
break into the tired brain. J Exp Soc Psychol 2009;45:320–6.
http://doi.org/10.1016/j.jesp.2008.10.001
Bettman JR, Luce MF, Payne JW. Constructive consumer choice processes. J Cons Res 1998;25:187–217. http://doi.org/10.1086/209535
Capa RL, Cleeremans A, Bustin GM, et al. Long-lasting effect of
subliminal processes on cardiovascular responses and performance. Int J Psychophysiol 2011;81:22–30. http://doi.org/10.1016/
j.ijpsycho.2011.04.001
Chen JCW, Li W, Lui M, et al. Left-frontal brain potentials index
conceptual implicit memory for words initially viewed subliminally. Brain Res 2009;1285:135–47. http://doi.org/10.1016/j.
brainres.2009.05.085
Chong TTJ, Husain M, Rosenthal CR. Recognizing the unconscious. Curr Biol 2014;24:R1033–5. http://doi.org/10.1016/j.cub.
2014.09.035
Chun MM, Phelps EA. Memory deficits for implicit contextual information in amnesic subjects with hippocampal damage. Nat
Neurosci 1999;2:844–7. http://doi.org/10.1038/12222
Daltrozzo J, Claude L, Tillmann B, et al. Working memory is partially preserved during sleep. PLoS ONE 2012;7:e50997. http://
doi.org/10.1371/journal.pone.0050997
Dew ITZ, Cabeza R. The porous boundaries between explicit and
implicit memory: behavioral and neural evidence. Ann NY
Acad Sci 2011;1224:174–90. http://doi.org/10.1111/j.1749-6632.
2010.05946.x
Dienes Z. Understanding Psychology as a Science: An Introduction to
Scientific and Statistical Inference. New York: Palgrave Macmillan,
2008.
Dienes Z. Using Bayes to get the most out of non-significant results. Quant Psychol Meas 2014;5:781. http://doi.org/10.3389/
fpsyg.2014.00781
7
Dijksterhuis A, Bos MW, Nordgren LF, et al. On making the right
choice: the deliberation-without-attention effect. Science
2006;311:1005–7. http://doi.org/10.1126/science.1121629.
Dijksterhuis A, Nordgren LF. A theory of unconscious thought.
Perspect Psychol Sci 2006;1:95–109. http://doi.org/10.1111/j.17456916.2006.00007.x.
Duss SB, Oggier S, Reber TP, et al. Formation of semantic associations between subliminally presented face-word pairs.
Conscious Cogn 2011;20:928–35. http://doi.org/10.1016/j.concog.
2011.03.018.
€ nggi J, et al. Unconscious relational encodDuss SB, Reber TP, Ha
ing depends on hippocampus. Brain 2014;137:3355–70. http://
doi.org/10.1093/brain/awu270.
Farooqui AA, Manly T. Anticipatory control through associative
learning of subliminal relations: invisible may be better than
visible. Psychol Sci 2015;26:325–34. http://doi.org/10.1177/
0956797614564191.
Ferrand L. The masked repetition priming effect dissipates when
increasing the inter-stimulus interval: evidence from word
naming. Acta Psychologica 1996;91:15–25. http://doi.org/10.1016/
0001-6918(95)00010-0.
Forster K, Booker J, Schacter DL, et al. Masked repetition priming:
lexical activation or novel memory trace? Bull Psychon Soc
1990;28:341–5. http://doi.org/10.3758/BF03334039.
Gaillard R, Cohen L, Adam C, et al. Subliminal words durably affect neuronal activity. NeuroReport 2007;18:1527–31. http://doi.
org/10.1097/WNR.0b013e3282f0b6cd.
pez J, Matas A, et al. “2 x 3” primes namGarcıa-Orza J, Damas-Lo
ing “6”: evidence from masked priming. Atten Percept
Psychophys 2009;71:471–80. http://doi.org/10.3758/APP.71.3.471
Greene AJ, Gross WL, Elsinger CL, et al. An fMRI analysis of the
human hippocampus: inference, context, and task awareness.
J Cogn Neurosci 2006;18:1156–73. http://doi.org/10.1162/jocn.
2006.18.7.1156
Greene AJ, Gross WL, Elsinger CL, et al. Hippocampal differentiation without recognition: an fMRI analysis of the contextual
cueing task. Learn Mem 2007;14:548–53. http://doi.org/10.1101/
lm.609807.
Greenwald AG, Draine SC, Abrams RL. Three cognitive markers
of unconscious semantic activation. Science 1996;273:1699–702.
http://doi.org/10.1126/science.273.5282.1699
Greenwald AG, Klinger MR, Schuh ES. Activation by marginally
perceptible (“subliminal”) stimuli: dissociation of unconscious
from conscious cognition. J Exp Psychol-Gen 1995;124:22–42.
Hannula DE, Greene AJ. The hippocampus reevaluated in unconscious learning and memory: at a tipping point? Front Hum
Neurosci 2012;6. http://doi.org/10.3389/fnhum.2012.00080.
Hannula DE, Ranganath C. The eyes have it: hippocampal activity predicts expression of memory in eye movements. Neuron
2009;63:592–9. http://doi.org/10.1016/j.neuron.2009.08.025.
Harand C, Bertran F, La Joie R, et al. The hippocampus remains
activated over the long term for the retrieval of truly episodic
memories. PLoS ONE 2012;7:e43495. http://doi.org/10.1371/jour
nal.pone.0043495.
Hassin RR. Yes it can: on the functional abilities of the human
unconscious. Perspect Psychol Sci 2013;8:195–207. http://doi.org/
10.1177/1745691612460684.
Hassin RR, Ferguson MJ, Shidlovski D, et al. Subliminal exposure
to national flags affects political thought and behavior. Proc
Natl Acad Sci USA 2007;104:19757–61. http://doi.org/10.1073/
pnas.0704679104.
Henke K. A model for memory systems based on processing
modes rather than consciousness. Nat Rev Neurosci
2010;11:523–32. http://doi.org/10.1038/nrn2850.
8
| Ruch et al.
Henke K, Buck A, Weber B, et al. Human hippocampus establishes associations in memory. Hippocampus 1997;7:249–56.
http://doi.org/10.1002/(SICI)1098-1063(1997)7:3<249::AIDHIPO1>3.0.CO;2-G.
Henke K, Mondadori CRA, Treyer V, et al. Nonconscious formation and reactivation of semantic associations by way of the
medial temporal lobe. Neuropsychologia 2003;41:863–76. http://
doi.org/10.1016/S0028-3932(03)00035-6.
Henke K, Reber TP, Duss SB. Integrating events across levels of
consciousness. Front Behav Neurosci 2013;7:68. http://doi.org/10.
3389/fnbeh.2013.00068.
Hesselmann G, Moors P. Definitely maybe: can unconscious
processes perform the same functions as conscious processes? Conscious Res 2015;584. http://doi.org/10.3389/fpsyg.
2015.00584.
Holdstock JS, Mayes AR, Isaac CL, et al. Differential involvement
of the hippocampus and temporal lobe cortices in rapid and
slow learning of new semantic information. Neuropsychologia
2002;40:748–68. http://doi.org/10.1016/S0028-3932(01)00192-0.
Karpinski A, Yale M, Briggs JC. Unconscious arithmetic processing: a direct replication. Eur J Soc Psychol 2016; n/a-n/a. http://
doi.org/10.1002/ejsp.2175.
Karremans JC, Stroebe W, Claus J. Beyond Vicary’s fantasies:
the impact of subliminal priming and brand choice. J Exp
Soc Psychol 2006;42:792–8. http://doi.org/10.1016/j.jesp.2005.
12.002.
Kawakami N, Yoshida F. Perceiving a story outside of conscious
awareness: when we infer narrative attributes from subliminal
sequential stimuli. Conscious Cogn 2015;33:53–66. http://doi.
org/10.1016/j.concog.2014.12.001.
Kiesel A, Wagener A, Kunde W, et al. Unconscious manipulation
of free choice in humans. Conscious Cogn 2006;15:397–408.
http://doi.org/10.1016/j.concog.2005.10.002.
Klapp ST, Haas BW. Nonconscious influence of masked stimuli
on response selection is limited to concrete stimulus-response
associations. J Exp Psychol-Hum Percept Perform 2005;31:193–209.
http://doi.org/10.1037/0096-1523.31.1.193.
Lau HC, Passingham RE. Unconscious activation of the cognitive
control system in the human prefrontal cortex. J Neurosci
2007;27:5805–11.
http://doi.org/10.1523/JNEUROSCI.4335-06.
2007.
Levy BR, Pilver C, Chung PH, et al. Subliminal strengthening:
improving older individuals’ physical function over time
with an implicit-age-stereotype intervention. Psychol Sci
2014;2525:2127–35. http://doi.org/10.1177/0956797614551970.
Lowery BS, Eisenberger NI, Hardin CD, et al. Long-term effects of
subliminal priming on academic performance. BASP
2007;29:151–7. http://doi.org/10.1080/01973530701331718.
McClelland JL, McNaughton BL, O’reilly RC. Why there are complementary learning systems in the hippocampus and neocortex: insights from the successes and failures of connectionist
models of learning and memory. Psychol Rev 1995;102:419–57.
Mealor AD, Dienes Z. Conscious and unconscious thought in artificial grammar learning. Conscious Cogn 2012;21:865–74.
http://doi.org/10.1016/j.concog.2012.03.001.
Miller J. Measurement error in subliminal perception experiments: simulation analyses of two regression methods. J Exp
Psychol-Hum Percept Perform 2000;26:1461–77.
Monahan JL, Murphy ST, Zajonc RB. Subliminal mere exposure:
specific, general, and diffuse effects. Psychol Sci 2000;11:462–6.
http://doi.org/10.1111/1467-9280.00289.
Moscovitch M. Recovered consciousness: a hypothesis concerning modularity and episodic memory. J Clin Exp Neuropsychol
1995;17:276–90. http://doi.org/10.1080/01688639508405123.
Mudrik L, Breska A, Lamy D, et al. Integration without awareness:
expanding the limits of unconscious processing. Psychol Sci
2011;22:764–70. http://doi.org/10.1177/0956797611408736.
Mudrik L, Faivre N, Koch C. Information integration without
awareness. Trends Cogn Sci 2014;18:488–6. http://doi.org/10.
1016/j.tics.2014.04.009.
Mudrik L, Koch C. Differential processing of invisible congruent
and incongruent scenes: a case for unconscious integration. J
Vis 2013;13:24. http://doi.org/10.1167/13.13.24.
Negash S, Kliot D, Howard DV, et al. Relationship of contextual
cueing and hippocampal volume in amnestic mild cognitive
impairment patients and cognitively normal older adults.
J Int Neuropsychol Soc 2015;21:285–96. http://doi.org/10.1017/
S1355617715000223.
Newell BR, Shanks DR. Unconscious influences on decision making: a critical review. Behav Brain Sci 2014;1–19. http://doi.org/
10.1017/S0140525X12003214.
Nieuwenstein MR, Wierenga T, Morey RD, et al. On making the
right choice: a meta-analysis and large-scale replication attempt of the unconscious thought advantage. Judgm Decis Mak
2015;10:1–17.
Ocampo B. Unconscious manipulation of free choice by novel
primes. Conscious Cogn 2015;34:4–9. http://doi.org/10.1016/j.con
cog.2015.03.007.
Olsen RK, Moses SN, Riggs L, et al. The hippocampus supports
multiple cognitive processes through relational binding and
comparison. Front Hum Neurosci 2012;6:146. http://doi.org/10.
3389/fnhum.2012.00146.
Parkinson J, Haggard P. Subliminal priming of intentional inhibition. Cognition 2014;130:255–65. http://doi.org/10.1016/j.cogni
tion.2013.11.005.
Reber TP, DoLam A, Elger CE, et al. Intracranial EEG correlates of
implicit relational inference within the hippocampus.
Hippocampus 2016;26:54–66. http://doi.org/10.1002/hipo.22490.
Reber TP, Henke K. Integrating unseen events over time.
Conscious Cogn 2012;21:953–60. http://doi.org/10.1016/j.concog.
2012.02.013
Reber TP, Luechinger R, Boesiger P, et al. Unconscious relational
inference
recruits
the
hippocampus.
J
Neurosci
2012;32:6138–48.
http://doi.org/10.1523/JNEUROSCI.5639-11.
2012.
Reder LM, Park H, Kieffaber PD. Memory systems do not divide
on consciousness: Reinterpreting memory in terms of activation and binding. Psychol Bull 2009;135:23–49. http://doi.org/10.
1037/a0013974.
Reuss H, Kiesel A, Kunde W, et al. Unconscious activation of task
sets. Conscious Cogn 2011;20:556–67. http://doi.org/10.1016/j.
concog.2011.02.014.
Ruby P, Caclin A, Boulet S, et al. Odd sound processing in the
sleeping brain. J Cognitive Neurosci 2008;20:296–311. http://doi.
org/10.1162/jocn.2008.20023.
Ruch S, Koenig T, Mathis J, et al. Word encoding during sleep is
suggested by correlations between word-evoked up-states and
post-sleep semantic priming. Front Psychol 2014;5:1319. http://
doi.org/10.3389/fpsyg.2014.01319.
Schlaghecken F, Eimer M. Masked prime stimuli can bias “free”
choices between response alternatives. Psychon Bull Rev
2004;11:463–8. http://doi.org/10.3758/BF03196596.
Shanks DR. Learning: from association to cognition. Annu Rev
Psychol 2010;61:273–301. http://doi.org/10.1146/annurev.psych.
093008.100519.
Simonson I. In defense of consciousness: the role of conscious
and unconscious inputs in consumer choice. J Cons Psychol
2005;15:211–17. http://doi.org/10.1207/s15327663jcp1503_5.
Subliminal messages—effects on decision-making |
Sklar AY, Levy N, Goldstein A, et al. Reading and doing arithmetic
nonconsciously. Proc Natl Acad Sci USA 2012. http://doi.org/10.
1073/pnas.1211645109.
Squire LR, Zola SM. Structure and function of declarative and
nondeclarative memory systems. Proc Natl Acad Sci USA
1996;93:13515–22.
Strahan EJ, Spencer SJ, Zanna MP. Subliminal priming and persuasion: striking while the iron is hot. J Exp Soc Psychol
2002;38:556–68. http://doi.org/10.1016/S0022-1031(02)00502-4
Tononi G. An information integration theory of consciousness.
BMC Neurosci 2004;5:42. http://doi.org/10.1186/1471-2202-5-42.
Tulving E. Episodic memory: from mind to brain. Annu Rev
Psychol 2002;53:1–25. http://doi.org/Article
Vadillo MA, Kostopoulou O, Shanks DR. A critical review and
meta-analysis of the unconscious thought effect in medical
decision making. Conscious Res 2015;636. http://doi.org/10.3389/
fpsyg.2015.00636.
Van den Bussche E, Van den Noortgate W, Reynvoet B.
Mechanisms of masked priming: a meta-analysis. Psychol Bull
2009;135:452–77. http://doi.org/10.1037/a0015329.
van Gaal S, De Lange FP, Cohen MX. The role of consciousness in
cognitive control and decision making. Front Hum Neurosci
2012;6:121. http://doi.org/10.3389/fnhum.2012.00121.
9
Van Opstal F, de Lange FP, Dehaene S. Rapid parallel semantic
processing of numbers without awareness. Cognition
2011;120:136–47. http://doi.org/10.1016/j.cognition.2011.03.005.
Verwijmeren T, Karremans JC, Bernritter SF, et al. Warning: You
are being primed! The effect of a warning on the impact of subliminal ads. J Exp Soc Psychol 2013;49:1124–9. http://doi.org/10.
1016/j.jesp.2013.06.010.
Verwijmeren T, Karremans JC, Stroebe W, et al. The workings
and limits of subliminal advertising: the role of habits. J Cons
Psychol 2011;21:206–13. http://doi.org/10.1016/j.jcps.2010.11.
004.
Vlassova A, Donkin C, Pearson J. Unconscious information
changes decision accuracy but not confidence. Proc Natl Acad
Sci
USA
2014;111:16214–8.
http://doi.org/10.1073/pnas.
1403619111.
Weinberger J, Westen D. RATS, we should have used Clinton:
subliminal priming in political campaigns. Polit Psychol
2008;29:631–51.
http://doi.org/10.1111/j.1467-9221.2008.
00658.x.
Züst MA, Colella P, Reber TP, et al. Hippocampus is place of interaction between unconscious and conscious memories. PLoS
ONE 2015;10:e0122459. http://doi.org/10.1371/journal.pone.
0122459.